Moonshot AI Launches Kimi K3, a 2.8-Trillion-Parameter Open-Weight Model
Chinese lab Moonshot AI released Kimi K3, a ~2.8-trillion-parameter mixture-of-experts model with a 1-million-token context window, built for long-horizon coding and agentic work, with full weights due by July 27.
Moonshot AI released Kimi K3 on July 16, its new flagship model, making it available immediately through Kimi.com, the Kimi app, Kimi Code and the Kimi API, with the full open-weight release to follow by July 27.
What shipped
Kimi K3 is a mixture-of-experts model with roughly 2.8 trillion total parameters, built on a new architecture Moonshot calls Kimi Delta Attention (KDA), a hybrid linear-attention mechanism, combined with a technique the company describes as Attention Residuals. The model supports a 1-million-token context window and native visual understanding, and Moonshot is pitching it specifically at long-horizon coding and agentic work: sustaining multi-step engineering tasks with minimal human supervision, navigating large codebases, and coordinating terminal tools. Two variants launched alongside the base model — K3 Max for chat and agent tasks, and K3 Swarm Max for large-scale parallel processing. On the API, Moonshot is pricing Kimi K3 at $0.30 per million input tokens with caching, $3.00 without, and $15.00 per million output tokens.
Where it stands
Moonshot's own benchmarking, as reported by MarkTechPost, says Kimi K3 delivers frontier-level results across its evaluation suite, consistently outperforming other open-weight models it was tested against, while still trailing the top proprietary systems on the market. The open-weight release is staggered: the model is live for use through Moonshot's hosted products and API now, but the downloadable weights, license and configuration files are not expected until July 27.
Why it matters
Kimi K3 is the latest in a run of large, open-weight releases from Chinese labs this year that trade off narrowly behind the leading proprietary models from OpenAI and Anthropic while remaining freely available to self-host and fine-tune — a dynamic that continues to pressure Western labs' pricing and closed-weight strategies. The model's explicit framing around long-horizon coding and agentic workflows, rather than general chat, also underscores how frontier labs are now optimizing flagship releases specifically for autonomous agent use cases instead of treating agentic capability as a secondary feature.
Sources
AI-assisted reporting, overseen by the AgentsAI team. Spotted an error? Let us know.
More ai news
DeepMind's Hassabis Calls for a FINRA-Style Global Watchdog for Frontier AI
Google DeepMind CEO Demis Hassabis published a manifesto urging the U.S. to spearhead an independent standards body that would safety-test frontier models before release, arguing AGI is only a few years away.
US Confirms Nvidia H200 AI Chips Are Now Shipping to China, Calls Volume 'Trivial'
A top US Commerce Department official told Congress that Nvidia H200 AI chip shipments to China have begun under the revised export-control framework, but described the quantity so far as 'trivial.'
Apple Sues OpenAI, Alleging Coordinated Theft of Hardware Trade Secrets
Apple filed a federal lawsuit accusing OpenAI, its io Products hardware unit, and two former Apple employees of stealing confidential iPhone-related trade secrets to build OpenAI's own consumer AI devices.
OpenAI Takes GPT-5.6 Public After Weeks of US Government-Gated Access
OpenAI opened GPT-5.6's Sol, Terra, and Luna models to the general public on July 9, after the Commerce Department's CAISI cleared a wider release that had been restricted to government-approved customers since late June.